Cinematic title sequence built with GPT Image, Seedream, LTX and MiniMax-H3
magratheya_64 · reddit · 2026-10-30
The author shares UNACCOUNTED, a speculative opening title sequence for a fictional investigative crime thriller — surveillance systems, empty offices, transit spaces, and a city after midnight — using title sequences as a way to practice visual storytelling without producing a full film.
Workflow:
- GPT Image and Seedream for image generation and visual development
- LTX and MiniMax H3 for image-to-video and camera movement
- ComfyUI for local generation workflows
- Post-production for shot selection, editing, color grading, sound, and pacing
Iteration focused on keeping a consistent visual language across shots and controlling camera motion. The author also explores how much directing can happen through shot selection, framing, and editing rather than relying on what the video model generates, and asks the community whether it reads as a real film opening.
More from Multimodal
- One-person AI music video: Suno + Hermes Agent + ComfyUI on a single RTX 3090 — Ok-Wolverine-5020 · 2026-10-30
- Same girl every scene vs. a new face each run: what a character LoRA actually changes — wuraang9 · 2026-10-30
- AI design accidentally triggers a human renaissance: taste beats execution by 2028 — EXM7777 · 2026-10-30
- Voice agents have a moment: Rime Prism, Deepgram buys Daily, Phonic tops GPT-Live — rdesh26 · 2026-10-30
- Fal's H3 Max still unreleased 2 months later as Pika's H3 Flex overtakes it on the leaderboard — TomLikesRobots · 2026-10-30
- MiniMax H3 emotional depth test turns a vague prompt into a tearful golden retriever — Ded_Ghost_007 · 2026-10-30